Should We Learn Most Likely Functions or Parameters?

Shikai Qiu, Tim G. J. Rudner, Sanyam Kapoor, Andrew Gordon Wilson

Keywords: Function-Space Modeling, Function-Space Regularization, Maximum A Posteriori Estimation, Generalization

TL;DR: This paper investigates the implications of directly learning the most likely function, instead of the most likely parameters, implied by a model and the data.

Abstract: Standard regularized training procedures correspond to maximizing a posterior distribution over parameters, known as maximum a posteriori (MAP) estimation. However, model parameters are of interest only insomuch as they combine with the functional form of a model to provide a function that can make good predictions. Moreover, the most likely parameters under the parameter posterior do not generally correspond to the most likely function induced by the parameter posterior. In fact, we can re-parametrize a model such that any setting of parameters can maximize the parameter posterior. As an alternative, we investigate the benefits and drawbacks of directly estimating the most likely function implied by the model and the data. We show that this procedure leads to pathological solutions when using neural networks and prove conditions under which the procedure is well-behaved, as well as a scalable approximation. Under these conditions, we find that function-space MAP estimation can lead to flatter minima, better generalization, and improved robustness to overfitting.

Submission Number: 105